I think Yes on (4) - make the founders rich, because most of the value will come from incentivising people to have an impact. Hence (B) on (7) - you auction the shares. The main disadvantage is that the greater cost might dissuade investors, which I think is mitigated a bit by making the certs cheaper via (B) on (2).
Two points on (2):
i) i’m not convinced that you (an initial buyer) get to ultimately decide how impact certs are valued. You at most get to write something about the asset in the contract, but ultimately it is the buyers and sellers who will decide how to value it. I feel if they are grantmakers seeking to incentivise good work at a low cost, they may be disinclined from buying the impact that would have happened anyway (even absent their existence as a funder). One way that preference could be satisfied is to give each share a number. Funders will value the first shares most, because they are fully “counterfactual”, but if half of the value comes from a thing that the founder would have done anyway, then shares beyond halfway will be worth nothing.
ii) so far we’ve talked about what happens when a funder buys a cert from a founder. There’s also the question of what happens when the next guy buys the cert from the funder. Should you pay for the whole impact of the funder paying the founder, or only the impact of paying the funder (i.e. apply the solution recursively). The latter would arguably be a somewhat distinct solution (c).
Thanks for the link, which I had previously missed and which does contain some important considerations.
I’ve been assuming that the people who set up the first impact market will have the opportunity to affect the “culture” around certificates, especially since many people will be learning what they are for the first time after the market starts to exist, but I agree that eventually it will depend on what buyers and sellers naturally converge to.
One way that preference could be satisfied is to give each share a number. Funders will value the first shares most, because they are fully “counterfactual”, but if half of the value comes from a thing that the founder would have done anyway
I don’t understand this—if you need $1 million to do the project, isn’t the one millionth one-dollar share purchased doing just as much good as the first? I think I’m missing something about the scenario you’re thinking of.
Sorry, I could have been clearer. Suppose we want to choose (B) on (2). And I ask for retro funding for a painting I drew.
Let’s consider three cases:
a) it had $500 of impact, and I was 0% likely to make the painting anyway.
b) it had $1k of impact, 50% likely anyway
b) $2k of impact, 75% likely anyway.
The “obvious” solution is that in each case, I can sell some certs, say 100 for $5 ea.
Alternatively, we could say that in:
case
a) each cert 1-100 is worth $5
b) certs 1-50 are worth $10 ea, 51-100 $0
c) certs 1-25 are worth $20 ea, 26-100 worth $0.
The second solution allows the world to know how useful the painting/project was, though at the cost of some complexity.
The bottom-line I think is that the contract should be clear about which thing it purports to be doing.
I think the theoretically optimal way to distribute credit would be to compute Shapley values for each contributor (including the the founder) and distribute the funding accordingly. But this is of course usually not possible (because of limited information).
Whether you use Shapley values seems orthogonal to the question I’m asking—whether to price the shares non-uniformly, in order to estimate relative contributions.
I think Yes on (4) - make the founders rich, because most of the value will come from incentivising people to have an impact. Hence (B) on (7) - you auction the shares. The main disadvantage is that the greater cost might dissuade investors, which I think is mitigated a bit by making the certs cheaper via (B) on (2).
Two points on (2): i) i’m not convinced that you (an initial buyer) get to ultimately decide how impact certs are valued. You at most get to write something about the asset in the contract, but ultimately it is the buyers and sellers who will decide how to value it. I feel if they are grantmakers seeking to incentivise good work at a low cost, they may be disinclined from buying the impact that would have happened anyway (even absent their existence as a funder). One way that preference could be satisfied is to give each share a number. Funders will value the first shares most, because they are fully “counterfactual”, but if half of the value comes from a thing that the founder would have done anyway, then shares beyond halfway will be worth nothing. ii) so far we’ve talked about what happens when a funder buys a cert from a founder. There’s also the question of what happens when the next guy buys the cert from the funder. Should you pay for the whole impact of the funder paying the founder, or only the impact of paying the funder (i.e. apply the solution recursively). The latter would arguably be a somewhat distinct solution (c).
I tried to think about (2) a bit here - https://forum.effectivealtruism.org/posts/eb28pDHzZz2RWh9Fh/will-impact-certificates-value-only-impact. It hasn’t had much discussion, so probably I’m still missing a lot of considerations.
Thanks for the link, which I had previously missed and which does contain some important considerations.
I’ve been assuming that the people who set up the first impact market will have the opportunity to affect the “culture” around certificates, especially since many people will be learning what they are for the first time after the market starts to exist, but I agree that eventually it will depend on what buyers and sellers naturally converge to.
I don’t understand this—if you need $1 million to do the project, isn’t the one millionth one-dollar share purchased doing just as much good as the first? I think I’m missing something about the scenario you’re thinking of.
Agree on affecting culture.
Sorry, I could have been clearer. Suppose we want to choose (B) on (2). And I ask for retro funding for a painting I drew.
Let’s consider three cases: a) it had $500 of impact, and I was 0% likely to make the painting anyway. b) it had $1k of impact, 50% likely anyway b) $2k of impact, 75% likely anyway.
The “obvious” solution is that in each case, I can sell some certs, say 100 for $5 ea.
Alternatively, we could say that in: case a) each cert 1-100 is worth $5 b) certs 1-50 are worth $10 ea, 51-100 $0 c) certs 1-25 are worth $20 ea, 26-100 worth $0.
The second solution allows the world to know how useful the painting/project was, though at the cost of some complexity.
The bottom-line I think is that the contract should be clear about which thing it purports to be doing.
I think the theoretically optimal way to distribute credit would be to compute Shapley values for each contributor (including the the founder) and distribute the funding accordingly. But this is of course usually not possible (because of limited information).
Whether you use Shapley values seems orthogonal to the question I’m asking—whether to price the shares non-uniformly, in order to estimate relative contributions.